PRE-SHIFT

Training that starts with your menus.

Pre-Shift is a training compiler. It ingests your food menus, wine lists, cocktail specs, and service documents — then generates structured, grounded training that adapts to each staff member and improves with every interaction. Every question traces to a real menu item. Every answer makes the system smarter.

How it works

1. Upload your menus and documents

Food menus, cocktail specs, wine lists, service standards, training manuals — PDF, URL, or manual entry. Pre-Shift extracts every item, enriches it with tasting notes, pairing suggestions, and service context, and builds a searchable knowledge base unique to your restaurant.

2. Generate grounded training

The system compiles your knowledge base into six training modalities — structured modules, timed assessments, adaptive diagnostics, Build a Lesson, Challenge Mode, and Ask Pre-Shift. Every question is generated from real menu items. Nothing generic. Nothing hallucinated.

3. Measure, adapt, improve

Staff answers feed into an ELO-based mastery system that tracks knowledge down to granular topics. The system identifies weak areas, generates targeted practice, and runs a weekly calibration cycle. The longer you use it, the better it trains.

See it in action

Source: 1975 Martini — Gin, Vermouth de Chambery, Orange Bitters

What distinguishes the vermouth used in the 1975 Martini from standard dry vermouth?

A. Bianco-style vermouth
B. Vermouth de Chambery, a lighter alpine style
C. Extra-dry vermouth
D. Sweet red vermouth

Every question traces to your menu

No generic quiz banks. Every question is generated from real menu data with guaranteed grounding.

Staff answer:

“It’s a classic gin martini with a lighter vermouth. Stirred, served in a coupe.”

75%
Semantic evaluation
Correctly identified lighter vermouth style
Proper technique: stirred, served in coupe
Missing: Chambery origin, alpine herb character

Grading that understands intent

Partial credit, rubric-based evaluation, and structured feedback — not just right or wrong.

Staff ELO Trend+47 this week
Cocktails85%
Wine62%
Food Menu91%
Weak area detected: Wine → Challenge generated

The system learns how to train your staff

ELO ratings, topic mastery tracking, and calibration algorithms that improve weekly.

Six training modalities

Every modality draws from the same canonical knowledge base — your menus, your documents, your restaurant's identity.

Modules

Structured curriculum by category — food, cocktails, wine, service. Staff works through questions one at a time with immediate feedback, citations, and ELO-adjusted difficulty.

Assessments

Admin-created evaluations that measure existing knowledge. Timed, no feedback during the test. Semantic grading with structured per-question feedback after submission. Grade override for managers.

Diagnostic

A 15-question adaptive assessment using computerized adaptive testing. Starts near the staff member’s current level and adjusts in real time — correct answers raise difficulty, wrong answers lower it.

Build a Lesson

Staff types a topic (“Italian wines,” “dessert pairings”) and the system generates a personalized lesson from the restaurant’s knowledge base in real time. Three teaching sections with inline comprehension checks, followed by a five-question assessment.

Challenge Mode

Twenty questions targeting the staff member’s weakest areas. The system pulls from granular topic mastery data to focus on exactly what each person needs to practice. Always 50/50 MCQ and short answer.

Ask Pre-Shift

A Q&A assistant that answers any question about your menu, grounded in your restaurant’s actual data. “What’s in the 1975 Martini?” “Which wines pair with the lamb?” Fast, cited, accurate.

The closed loop

The closed loop is what makes Pre-Shift a compounding system, not a static one.

Every week, a calibration agent analyzes staff performance data — identifying coverage gaps, flagging low-quality questions, mining common misconceptions from real wrong answers, and finding the highest-performing questions to use as generation examples. That analysis produces a generation brief that the system consumes the next time it creates content — injecting real exemplars, pre-loading discovered misconceptions into rubrics, and prioritizing undertested topics. A separate quality audit retires questions that don't discriminate well or that staff consistently flag.

The result: the training your team takes in month three is materially better than the training they took in month one — automatically, without anyone manually updating a quiz.

Ready to see it in action?

See how Pre-Shift turns your menus into training your team actually uses.